On Tue, Mar 23, 2021 at 12:03:36PM -0400, Bruce Momjian wrote:
> On Tue, Mar 23, 2021 at 05:21:30AM +0000, PG Bug reporting form wrote:
> > postgres=# set intervalstyle='postgres';
> > SET
> > postgres=# select interval '-1 year -1 day';
> > interval
> > ------------------
> > -1 years -1 days
> > (1 row)
> >
> > Expected output:
> > -1 year -1 day
> >
> > The code
> > (https://github.com/postgres/postgres/blob/master/src/backend/utils/adt/datetime.c#L4193)
> > pluralizes a time unit if the value is not 1, should check both -1 and 1.
From the first several links at
https://www.google.com/search?q=negative+one+singular+plural the choice is a
matter of dispute among English grammar commentators. Singular "-1" is
popular, and plural "-1" is also popular.
> Wow, interesting. Seems there were a few places in the code that
> handled this properly, and several that didn't. The attached patch
> fixes these and updates the regression tests. I will apply this for PG
> 14 in the next few days. This is too likely to break something to
> backpatch. Thanks for the report.
Let's not change from one popular spelling to another when doing so creates a
compatibility hazard. That is to say, I think PostgreSQL would be better with
this patch reverted.
If we did want to standardize on singular for -1, EVALUATE_MESSAGE_PLURAL()
would be a key bit of code to change.